A token that has just run up and just got loud usually gives it back.

Three signal streams, one rule

The streams differ in one number: how far the price had to run before the scanner takes interest. Signal 1 has the lowest bar and fires most often, Signal 3 the highest and fires rarely. The deeper the run-up behind a signal, the deeper the fall that typically follows.

The record starts on 26 July 2026. Up to 11 September 2026 it is today's rules run over the data we had collected; from 11 September every entry is a live signal, decided once, the hour it closed. It covers one stretch of market and carries no promise about the next one.

A great deal of crypto attention is paid for

Bought enthusiasm is stripped out by our own algorithms before it can reach a signal. The filter does not ask who is speaking. It asks whether the talking looks like a conversation or like a campaign: whether the sources are independent, whether anyone answered, whether the engagement fits the audience, and whether it lasted past the first day. One account posting ten times counts as one account.

A third condition holds the whole thing back when the market itself is surging. On those days overheated tokens keep climbing with everything else, and short positions lose money, so the scanner waits.

Every position is run the same way, and printed

A short opens at the signal price and closes exactly 72 hours later, whatever the result at that moment. No stop losses, no early exits. The number on each row is the full 72-hour outcome measured against the market, and the losing rows sit next to the winning ones, sorted by time and by nothing else. A separate program recounts every position from the raw signal record each hour and shouts when the screen and the record disagree.

Trading costs are not in these figures: exchange fees, the funding paid on a short, and slippage are real and are measured separately.

Free and subscription

The record is public. The live signals are what you pay for.

Checkout